Mobile UA Tracking, Budgeting & Experimentation — the Full Picture

“CVR moved 0.7% → 0.9%. Was it the new subtitle?” That’s a real question from an ASO thread last week, and it’s not a one-off. Ask any mobile UA manager running more than two channels why they can’t answer basic questions about their own campaigns, and you’ll hear a version of the same story: attribution is broken in one tool, the budget lives in three spreadsheets, and nobody remembers what changed last Tuesday that moved the numbers.

This guide is the one place to understand how mobile UA tracking, budgeting, and experimentation fit together — not as three separate tools, but as one system. If you’re managing spend across Google, Meta, TikTok, Apple Search Ads, and one or two DSPs, this is the map.

What Mobile UA Tracking Actually Means

Most teams use “tracking” as a catch-all word for three different things that break in three different ways.

Tracking vs. Attribution vs. Budget Planning

• Tracking = raw event data (installs, in-app events) flowing from your app to your analytics or MMP.
• Attribution = deciding which touchpoint — which ad, which channel — gets credit for a conversion.
• Budget planning = deciding how much to spend where, based on what tracking and attribution tell you.

Most breakdowns happen at the seams between these three, not inside any single one of them.

Why Stitching Together Separate Tools Creates a Gap Between Analysis and Action

Ask this in any UA community and you’ll get a version of the same complaint: how do you close the gap between analyzing your data and actually executing on it? The data says something. Nobody translates that into the next campaign decision, because there’s no place that connects “here’s what changed” to “here’s what we did about it.”

That missing layer — call it a decision log — is the difference between a team that reacts to dashboards and one that improves week over week. It sits between your raw tracking (MMP, ad platform data) and your actual campaign changes.

Attribution & Tracking Foundations

What Your MMP Is Actually Telling You (and When It Isn’t)

A Mobile Measurement Partner — AppsFlyer, Adjust, Kochava — is still the backbone of cross-channel attribution. No spreadsheet replaces the SDK-level event tracking it does. But an MMP can report zero attribution even when your events clearly show up in a channel’s own dashboard. The usual suspects: an ATE (App Tracking Transparency) flag not set correctly, a broken postback URL, or a campaign structure that doesn’t match what the MMP expects. One misconfigured flag can make four weeks of spend look unattributed even though the installs happened.

The Store Listing Blind Spot

“CVR moved 0.7% → 0.9%. Was it the new subtitle?”

Nobody can answer that question if title, subtitle, and screenshots change without any version log. You end up unable to attribute a CVR shift to a specific change — was it the subtitle, the icon test, or seasonality? Without a lightweight version control layer around your store listing, every CVR movement becomes a guess.

When a Spreadsheet Tracker Beats (or Complements) an MMP

An MMP tells you what happened at the event level. It rarely tells you why you changed your bid caps last Tuesday, or what you expected a creative refresh to do — MMPs simply aren’t built to log decisions. A lightweight tracker sitting next to your MMP, logging changes and expected outcomes, closes that gap without replacing the attribution layer underneath it.

Channel Performance — Reading Benchmarks Before You Optimize

Before touching bids or creative, you need a baseline. Below is a snapshot; the full breakdown — including retention and ROAS benchmarks by monetization model — is in the free UA Channel Scorecard.

Channel	CPI (Android)	CPI (iOS)	Notes
Apple Search Ads	—	$2.96	Lowest CPI among iOS channels
Meta	$1.11 blended	$1.11 blended	D7 retention 6% below network avg
Hypercasual (all)	$0.30–$6.00	$0.50–$12.00	Genre-dependent, varies by competition
Strategy / Midcore	~$4.00	~$5.50	Higher CPI, higher monetization ceiling

If your numbers are meaningfully off these ranges, that’s the first thing to investigate — not your bidding strategy.

The Apple Search Ads Bottleneck

“CPAs shift overnight… you’re reacting to yesterday’s data.”

At real scale, Apple Search Ads becomes a manual optimization problem. Budget caps get set the night before and are already wrong by mid-morning. Bid changes take hours to reflect in reporting. This is less about picking the wrong bidding strategy and more about the fact that manual optimization simply can’t keep pace with how fast the auction moves — a pattern that shows up constantly wherever teams debate trusting automated bidding versus doing it by hand.

The DSP Honeymoon Cliff

“honeymoon cliff after ~D7… panic-buying cheap installs from RU/CIS.”

A pattern worth watching if you run AppLovin, Liftoff, or Vungle at scale: strong performance in the first week, then a drop-off. The DSP runs out of its best inventory, and to keep hitting spend targets, starts filling the gap with lower-quality traffic — often geographically concentrated in regions with weaker unit economics. If your D1 numbers look great and your D30 numbers don’t, this is one of the first things worth checking.

Discoverability Isn’t Just a UA Problem

Not every underperformance is a targeting problem. Some apps sit published for years with close to zero organic downloads and no clear diagnostic path — a version of “still didn’t get downloads, I don’t know why” shows up constantly in Google Play communities. Before scaling paid spend into a listing, it’s worth ruling out a structural discoverability issue first: keyword indexing, category placement, or a listing that isn’t converting impressions into store visits.

Budgeting for Mobile UA

Why DIY Spreadsheet Dashboards Usually Fall Apart

The pattern is consistent: a founder or UA manager builds a budget tracker in Sheets, it works for a few weeks, then it quietly stops being updated. Three failure points show up over and over: no version control, so nobody can tell what changed; no link between a budget change and the result it produced; and manual data entry that falls behind the moment more than one person touches the sheet.

A Quick Self-Audit

Before building — or buying — anything, it’s worth answering a short list of questions honestly:

• Can you tell, right now, which channel is actually profitable this month?
• If your CPA moved 20% last week, could you say why within five minutes?
• Is there one place — not four — where budget, actuals, and decisions live together?

If the answer to any of these is no, that’s the gap worth closing first, regardless of which tool ends up filling it.

Experiments & Decision Logs — Turning Insight Into Action

Why a BI Dashboard Alone Doesn’t Solve This

A BI tool like Looker or Tableau is excellent at showing you what happened. It has no concept of “what we changed” or “what we expected to happen” — that context lives in someone’s memory, a Slack thread, or nowhere at all. That’s the real gap behind the recurring question of closing the loop between analysis and execution.

What a Minimum Decision Log Needs

Three fields, at minimum: what changed — bid, creative, budget cap, listing; what you expected; and what actually happened. It sounds almost too simple to matter, but most teams don’t have even this much structure — which is exactly why the same mistakes get repeated every quarter.

A Blind Spot in A/B Testing Data

“only getting data from 10% of the qualifying audience.”

Even when a team runs structured experiments, the sample can be smaller than it looks. A commonly reported issue with Firebase Remote Config is exactly this. Whether it’s an expected sampling limitation or a config problem, the practical effect is the same — a test that looks conclusive might be running on a tenth of the population you think it is. Worth checking before trusting any A/B result at face value.

Build vs. Buy vs. BI Tool — Choosing the Right Layer

Three broad paths, and they’re not mutually exclusive:

For most teams under a certain scale, the second option is the pragmatic middle ground — the structure of a system, the flexibility of a spreadsheet, without the sprawl of a full DIY build. The Mobile UA Budget Tracker was built around exactly that gap: a Setup tab, Monthly Budget Plan, Performance Tracker, Dashboard, and Experiment Log in one file, no integration required.

FAQ

How is mobile UA tracking different from an MMP?

An MMP handles attribution at the event level — which install came from which ad. Tracking, in the broader sense used here, includes budget planning and a decision log layered on top of that data. You need both; they answer different questions.

How much budget do I need before proper tracking is worth setting up?

Structure matters more than spend size. A team spending $500/month with no way to trace what changed what is worse off than a team spending $5,000/month with a basic decision log. Set the system up before you scale spend, not after.

Can a spreadsheet tracker replace AppsFlyer or Adjust?

No — and it shouldn’t try to. An MMP does SDK-level attribution a spreadsheet can’t replicate. A tracker’s job is the layer above that: budget, benchmarks, and the decision context an MMP was never built to hold.

How often should channel benchmarks be reviewed?

Monthly at minimum, weekly if you’re actively scaling a channel. Benchmarks like CPI and retention shift with seasonality, competition, and platform changes — a number that was normal in Q1 can be a red flag by Q3.

Where to Start — Three Steps

1. Download the UA Channel Scorecard and check your current numbers against it.

2. Run the UA Tracking Audit Checklist to see exactly where your system has gaps.

3. If the audit points to “no central system,” take a look at the Mobile UA Budget Tracker — see the full six-tab breakdown.